NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 Contents lists available at SciVerse ScienceDirect NJAS - Wageningen Journal of Life Sciences jou rna l h omepage: www.elsev ier .co Incentives for cocoa bean production in Ghana: D W. Quarm tec, a Development he Net b College of Agr c Institute of St d Laboratory of a r t i c l Article history: Received 21 Se Accepted 1 Jun Available onlin Keywords: Value chain Information asymmetry Institutional experiments al fac llecte lti-sta ing co nienc t dist quality is important to all categories of actors in the cocoa sector, interactions among them are hampered by problems of information asymmetry that result especially in farmers evading recommended practices. While cocoa sector policies ensure the export of premium quality cocoa, policies have not sufficiently alleviated the information problem especially in the relation between farmers and cocoa purchasing agents. It explains why Ghanaian farms have not been able to reach their full potential to produce more than 1,000,000 metric tons of premium quality cocoa annually. Amongst other options, self-selection 1. Introdu Cocoa be premium co quality pre from cocoa enue and ab quality coc measures [ increased if harvested c The que their full y tutional fa Abbreviatio – Strengthenin buying compa ∗ Correspon sity, P.O. Box 8 Tel.: +31 233 2 E-mail add 1 Bank of Gh 1573-5214/$ – http://dx.doi.opolicies, such as quality testing with price premiums, are recommended for testing as potential incentive mechanisms that address information asymmetry. © 2012 Royal Netherlands Society for Agricultural Sciences. Published by Elsevier B.V. All rights reserved. ction ans exported from Ghana attract a substantial quality mpared with cocoa from other countries [1,2]. These miums partly explain the high revenue Ghana earns , amounting to about 30% of Ghana’s total export rev- out 4% of GDP.1 Ghana’s status as a supplier of premium oa is a result of strict post-production quality control 3]. The volume of high-quality cocoa beans could be farmers could be motivated to enhance the quality of ocoa beans [4–7]. stion is why cocoa farms in Ghana do not reach ield and quality potential. It is thought that insti- ctors restrain farmers’ incentives to enhance the ns: COCOBOD, Ghana Cocoa Board; CoS-SIS, Convergence of Sciences g Innovation Systems; GDP, gross domestic product; LBC, licensed ny; PPRC, Producer Price Review Committee. ding author at: Development Economics Group, Wageningen Univer- 130, NL-6700 EW Wageningen, The Netherlands. 0 2749200. ress: qandq2002@hotmail.com (W. Quarmine). ana Annual Report, 2009. quality of the cocoa beans they produce [8]. The Convergence of Sciences – Strengthening Innovation Systems (CoS-SIS) research programme (see Introduction to this issue), of which this study forms part, proposes to tackle the quality concern in Ghana’s cocoa sector through experimenting with institutional change [9–13]. The CoS-SIS approach involves identification and facilitation of institu- tional changes that might provide incentives for Ghanaian cocoa farmers to enhance the quality of the cocoa beans they produce. Quality cocoa here refers to cocoa that is well fermented, dried, and free from disease, contamination and other physical defects. Because the kinds of change that might achieve this quality objective cannot be known in advance, industry stakeholders acting together in a Concertation and Innovation Group (CIG) have been convened to identify, develop and implement institutional exper- iments to discover which options work best. In the cocoa domain, the success or otherwise of the CoS-SIS approach will depend on how thoroughly the issue of farmers’ incentives to enhance cocoa bean quality is understood. This paper is based on a diagnostic study of the institutional factors that have been identified as constrain- ing farmers’ practices to enhance cocoa bean quality. Other studies that analyse the cocoa sector of Ghana from an institutional point of view do not pay much attention to the incentives to sustain quality by cocoa farmers [14–18]. see front matter © 2012 Royal Netherlands Society for Agricultural Sciences. Published by Elsevier B.V. All rights reserved. rg/10.1016/j.njas.2012.06.009inea,b,∗, R. Haagsmaa, O. Sakyi-Dawsonb, F. Asan Economics Group, Wageningen University, P.O. Box 8130, NL-6700 EW Wageningen, T iculture and Consumer Sciences, University of Ghana, Legon, Ghana atistical Social and Economic Research, University of Ghana, Legon, Ghana Entomology, Wageningen University, The Netherlands e i n f o ptember 2011 e 2012 e 16 September 2012 a b s t r a c t This paper investigates the institution of cocoa beans in Ghana. Data were co village, district, and national level. Mu 12 purchasing agents of licensed buy Wasa Akropong cocoa districts. Conve organizations and service providers am/locate /n jas oes quality matter? A. van Huisd, D. Obeng-Oforib herlands tors that constrain farmers’ incentives to enhance the quality d at three levels of aggregation in the cocoa bean value chain: ge cluster sampling was employed to sample 120 farmers and mpanies from 12 villages in Assin Foso, Suhum, Dormaa and e sampling was used to sample key informants from relevant rict and national levels. The study revealed that, even though 8 W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 The objective of this paper is to identify the institutional factors that act as a disincentive to farmers to enhance the quality of their cocoa bean How do the What is the institutiona quality prob to address t The stud cocoa bean interactions tions shape Institutions practices, r tions betwe refer mainl the physica They includ buy cocoa b the Cocoa B governs the 2. Method 2.1. Study c CoS-SIS because of t lem identifi study was c sector [8]. S the cocoa b incentives the third ph study of the from the p appropriate improveme performanc 2.2. Sampli The stud part of Gha at three lev Multi-stage at the villag collected fr the district Control Com and the Dis used to sele at the natio input comp In order 38 districts based on th quality [22] zone: Assin deciduous est zone, an simple ran cocoa-grow tricts. Nkra Assin Foso Table 1 Districts and villages sampled in this study. ical zo l sava ous r rest tional ield d hum a Ah from o-s illag selec discu edge tive pling usin of 1 as ob ers o ata co mi-s 0 farm s an r inf ents quen lyse t tativ cio-t cal an lped iolog n to ed [ find mee Key er 20 e stu ny, G (LBC na, a ults efinition and perceptions of cocoa bean quality international cocoa market defines quality in four main as applied and certified in exporter–buyer contracts: (1) al quality; (2) bio-chemical quality; (3) process quality; and in quality [24,25]. Physical quality sical quality relates to moisture content, disease infesta- efectiveness of beans, mouldiness, and the presence of matter [26,27]. Both the domestic and the internationals. Specifically, the paper addresses four questions: (1) key actors in Ghana’s cocoa sector define quality? (2) state of cocoa bean quality in Ghana? (3) What are the l and socio-technical reasons underlying the cocoa bean lem? (4) What institutional or policy options are likely he quality problem in Ghana’s cocoa? y was based on two assumptions: (1) the quality of s produced and exported depends on the actions and of all the actors in the cocoa sector and (2) institu- the incentives for these actions and interactions [19]. are “. . .the set of common habits, routines, established ules or laws that regulate the relations and interac- en individuals and groups” [20,21]. Actors in this study y to those individuals or organizations involved with l handling of cocoa beans from production to export. e farmers, licensed buying companies (LBCs), which eans from farmers on behalf of the third-party actor, oard (COCOBOD). The latter is a parastatal entity that industry and also handles all cocoa bean exports. ology ontext selected the cocoa sector as one of its research domains his sector’s importance to the national economy. Prob- cation was carried out in three phases. First, a scoping onducted that identified the main concerns in the cocoa econd, stakeholder workshops were held throughout elt to identify and prioritize the possibly inadequate for farmers to enhance quality. This paper relates to ase, and reports the findings of a follow-up diagnostic prioritized problem. The analysis is conducted mainly erspective of the cocoa farmer. Such a perspective is since it is the farmers’ response to any institutional nt that is likely to enhance the quantity and quality e of the sector. ng procedures y was carried out in the cocoa districts of the southern na from June to September 2010. Data were collected els of aggregation: village, district, and national level. cluster sampling was employed to select respondents e and cocoa district levels. At the village level, data were om farmers and the purchasing clerks of the LBCs. At level, data were obtained from the staff of the Quality pany, the Cocoa Extension Co-ordinators of COCOBOD, trict Officers of the LBCs. Convenience sampling was ct one key informant from each relevant organization nal level, including COCOBOD, cocoa processors, and anies. to select respondents from village and district levels, the were clustered into four cocoa agro-ecological zones, e assumption that climatic factors affect cocoa bean . One cocoa district was selected randomly from each Foso from the coastal savannah zone, Suhum from the rain forest zone, Wassa Akropong from the rain for- d Dormaa Ahenkro from the transitional zone. Next, dom sampling techniques were used to select three ing villages from a list of villages in each of the four dis- nfuom, Ayitey, and Wura Kesse were selected from the district; Anum Asuogya, Duodukrom, and Kuano from Ecolog Coasta Decidu Rain fo Transi Source: F the Su Dorma Valley A tw each v sively group knowl a tenta of sam village sample tion w memb 2.3. D A se the 12 cussion Furthe docum ing fre to ana quanti The so techni tool he vide b going o identifi The during (CIG). [Octob into th Compa Kokoo of Gha 3. Res 3.1. D The ways, physic (4) orig 3.1.1. Phy tion, d foreignne District Villages nnah Assin Foso Nkranfuom Ayitey Wurakesse ain forest Suhum Asuogya Duodukrom Kuano Wassa Akropong Oppong Valley Bogoso Donkor Krom zone Dormaa Ahenkro Esikeso Diabaa Nkrankwanta ata 2010–2011. District; Nkrankwanta, Esikesu, and Diabaa from the enkro District; and Bogoso, Donkor Krom, and Oppong the Wassa Akropong District (Table 1). tage sampling procedure was used to select farmers in e. In the first round, five cocoa farmers were purpo- ted in each village and invited to participate in focus ssions. They were selected on the basis of their general of the sampled communities, and helped us to draw up list of cocoa farmers in the village. In the second round , 10 cocoa farmers were randomly selected from each g the tentative list as sampling frame, making a total 20 farmers. In addition, in each village further informa- tained from two purchasing clerks of the LBCs and two f the government’s mass cocoa spraying gangs. llection and analytical procedures tructured questionnaire was used to collect data from ers. A checklist was used to guide the focus group dis- d key informant interviews with the institutional actors. ormation was obtained from a desk review of official supplied by COCOBOD. Descriptive statistics involv- cies and percentages and content-analyses were used he quantitative and qualitative data, respectively. The e data were analysed using SPSS (version 17, SPSS Inc.). echnical root system tool [23] was used to analyse the d institutional causes of poor cocoa bean quality. This us first to identify the central problem and then to pro- ical or technical explanations for the problem, before unravel the institutional cause of the technical reasons 23]. ings from the diagnostic study were further validated tings of the cocoa Concertation and Innovation Group findings were presented also at a meeting of the CIG 10] where participants were invited to make their input dy, attended by representatives from the Quality Control hana Standards Board, Cocoa Inputs Company, Kuapa ), Cocoa Research Institute of Ghana (CRIG), University nd the Ministry of Finance. and analysis W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 9 Table 2 Percentages of farmers who agree with the statement: “Bean quality is important for Ghana’s cocoa sector”. o (n = 30) Wassa Akropong (n = 30) Total (n = 120) Strongly disa 0.0 5.8 Disagree 33.3 21.7 Neutral Agree Strongly agr Source: Field d market enf ier to assess physical qu market stan of the likel sit between standards, beans are w than 7.5% a Grade II co standards t in addition ture conten of uniform have accep the farmers requiremen 3.1.2. Bio-c Bio-chem icals, heavy residues lef tion of coco concerns ha On two occ Japanese an chemical re the parame unknown t interviews r tance of che usage in th ting up labo cocoa beans 3.1.3. Proce Biochem i.e., whethe child labou subsequent trade) [29]. sider proce quality. How good qualit steps to in instance, ch mized. Som zones, whil LBC. In gener that the m importance tor (Table 2 bean qualit farmers and cilita 2 ate o en th ial of coc 0 MT the o pu [7]. N e acc on fa purch ies to ther ts th reas he 20 ords s no r co rmer they stud fficia omet ict d e the ve. If ns to d bea d co repo of pu Com stan an 10 prem with e 3, d purc here le sh e of c o beSuhum (n = 30) Dormaa (n = 30) Assin Fos gree 10.0 3.3 10.0 20.0 13.3 20.0 0.0 3.3 0.0 10.0 46.7 10.0 ee 60.0 33.3 60.0 ata 2010–2011. orce physical quality standards because they are eas- prior to export. COCOBOD sets and enforces minimum ality standards that are higher than the international dards. These higher standards are imposed because ihood that the cocoa beans will deteriorate in tran- farms and the final market destination. By Ghanaian a bag of cocoa beans is classified Grade I cocoa if the ell fermented, have a moisture content not higher nd do not contain more than 3% beans with defects. coa is comparable to international premium quality hat accept 4–8% beans with any of the other defects, to good fermentation, and not more than 8.5% mois- t. Moreover, all cocoa bags must contain cocoa beans size. While all other actors in Ghana’s cocoa sector ted the physical quality standards of the COCOBOD, interviewed were generally unaware of these specific ts. hemical quality ical quality focuses on butter content, flavour chem- metals, toxic compounds, and the level of chemical t on the beans [28]. Ghana is known for the produc- a beans of a high chemical quality. Recently, however, ve been raised about chemical residues on its beans. asions cocoa beans from Ghana have been rejected by d American markets because they exceeded maximum sidue requirements. With the exception of COCOBOD, ters of the chemical quality standards appear to be o most cocoa actors. Nevertheless, the key informant evealed that COCOBOD itself acknowledges the impor- mical quality and has taken steps to control chemical e cocoa sector. COCOBOD is also in the process of set- ratories to test for the presence of chemical residues on prior to export. ss quality ical quality refers to the production process of cocoa r organic or inorganic methods are employed; whether r is used; and whether the production process and rewards benefit the farmer and his community (fair The farmers and the LBCs interviewed did not con- ss quality an important component of cocoa bean ever, COCOBOD is interested in maintaining Ghana’s y image on the international market and has taken clude process-quality control into its policies. For ild labour on cocoa farms in Ghana has been mini- e cocoa districts have been marked as organic cocoa and fa cocoa.” 3.2. St Giv potent quality 750,00 tion of prior t waste provid waste waste activit tion ga sugges has inc 7% in t cial rec there i in othe that fa beans The BOD o LBCs s at distr becaus defecti the bea infeste license It is weeks Buying quality less th tional beans In Tabl beans size, w The tab volum small te Kuapa Kokoo Ltd. has been certified as a fair trade al, the results from the interviews with farmers revealed ajority of the respondents (71%) acknowledged the of cocoa bean quality to the development of the sec- ) and all of the LBC staff interviewed regarded cocoa y as being important to the sector. The perspective of LBCs is in line with COCOBOD’s vision to “Encourage domestic m beans was 25% per ann of light crop 2 Source: Th0.0 0.0 26.7 23.3 40.0 48.3 te the production and processing of premium quality f cocoa bean quality in Ghana e current area under production, cocoa farms have the producing up to 1,000,000 metric tons (MT) of premium oa annually, yet actual production has not exceeded [30]. One explanation for this is that a great propor- cocoa output suffers from diseases and poor handling rchasing by the LBCs, and are thus regarded as cocoa ational data on cocoa waste are available, but do not urate information on the total volume of cocoa going rms. This is because they capture only official cocoa ased by licensed buyers who do not extend their buying all cocoa-growing communities in Ghana. Yet, informa- ed from various editions of COCOBOD’s annual reports at cocoa waste as a percentage of annual production ed from about 1.5% in the 1999–2000 season to about 08–2009 season. Although these figures represent offi- of cocoa waste purchased in just a few communities, reason to expect that the figures will be very different mmunities and they can be taken as a clear indication s could do more to increase the volume of quality cocoa sell. y found, based on key informant interviews with COCO- ls and LBCs, that even the cocoa beans bought by the imes fail COCOBOD’s strict quality control procedures epots. Sometimes bags of cocoa beans fail quality tests beans are not well dried, not of uniform size, or simply the beans are not well dried, then LBCs are asked to dry the appropriate moisture content. Defective, small, or ns should either be thrown away or, if possible, sold to coa waste buyers. rted for instance that in the 2004–2005 season, after six rchases, only 15% of all cocoa purchased by the Produce pany (an LBC subsidiary of COCOBOD) met minimum dards [5]. In the 2005–2006 and 2006–2007 seasons, % of the cocoa purchased by LBCs could meet interna- ium quality standards because the percentage of cocoa a purple instead of chocolate colour was too high [31]. ata from COCOBOD for the different categories of cocoa hased are given. Main crop cocoa beans are bigger in as light crop and small beans are too small for export. ows that over the last 10 years at least 10% of the total ocoa beans purchased from farmers annually was too exported. The rejected beans are sold at a discount to anufacturers. The proportion of light crop and small higher between 2002–2003 and 2007–2008, averaging um. It is unclear what explains the very low percentage and small beans for 2008–2009. e mission statement of COCOBOD. 10 W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 Table 3 Proportion of light crop and small beans purchased in Ghana, 1999–2009. Season Main crop Light crop Small beans 1999–2000 87.92 10.50 1.58 2000–2001 84.27 10.86 4.87 2001–2002 88.67 10.28 1.05 2002–2003 76.71 18.83 4.46 2003–2004 71.15 26.59 2.26 2004–2005 68.68 28.41 2.90 2005–2006 70.13 28.25 1.62 2006–2007 81.07 11.44 7.49 2007–2008 82.45 9.91 7.64 2008–2009 98.25 1.58 0.16 Source: COCOBOD, unpublished data. 3.3. Main causes of the cocoa bean quality problem Fig. 1 gives a diagrammatic representation of what we think, after analysing responses from interviews, are important technical and institutional causes of the quality concerns mentioned in the previous section. 3.3.1. Technical explanation of the quality problem Our findings suggest that inappropriate pre- and post-harvest activities are the main technical causes of the quality problem. Table 4 presents the views of the sampled farmers on the practices that result in cocoa beans that cannot be marketed. Even though the choice of variety of cocoa planted affects the bio-chemical quality Consequences Incr eased cost of quali ty cont rol Forgo ne rev enues due to bean loss and quali ty discou nt Central prob lem Produ ctio n of low gr ade and unmarketa ble coco a be ans Biological causes of the problem Poo r farming practices Poor proc essing Choice of planting material Too lon g or too short pod storag e Bad pest and disease management Poor harve sti ng practices Bad pod breaking meth ods Bad fermentatio n prac tic es Inadequ ate drying an d pol ishing Calendar- rather th an need-based spraying Irregular harvesting Use of sha rp objec ts Not turning fermentation heap Fermentin g for less than 6 days Too large fermentation heap Instit ution al causes of the prob lem Off -season financial shortages Informat ion asymme try Inad equ ate knowledge  Unf avourable la nd use contracts  Produ cer pric e policy (tim eli ness of price ann ounce ment)  Rent see king beha viour of LBCs  High co st of lending  Theft of cocoa bea ns  Lack of farm monito  Lack of quali ty tes tin grading and pr ice differentiation at farmgate  We ak farmer organiz ati ons Poor ly covered ferment ati on heap Brea king overrip e and diseased Inadequ ate incenti ve Fig. 1. Biological and institutional causes of the sub-optimal qualit Source: Diagnostic Study, 2010/2011.ring g,  Inc ons istent exte nsion poli cy  Lo w ext ensio n coverage y performance of cocoa farmers. W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 11 Table 4 Percentages of farmers who agree that the listed practices result in poor bean quality. Farm practic Foso Variety type Poor farm sa Type of chem Frequency o Length of po Poor pod bre Poor fermen Inadequate d Source: Field d Table 5 Percentages of Price policy n Foso The price (am General pric Timing of an Cocoa bonus Source: Field d [32], only 7 mines the fi farmers ack appropriate tion (84%), beans they the necessa and inputs. because the 3.3.2. An in The mai that the dom tion asymm LBCs and C tices they ap cocoa bean tion about standards t the cocoa ch The asym do not have process of fa ities could e farmers’ pra spraying of ensure that ities, such a trees are sp pod). Prior to the farmer tices. The c cocoa grow organizatio during prol is an umbre Coffee Shea at national Another is that in G prior to pur beans are b depots that Mea mers ers rs thu king A kno m th as o n fa OD p ot re trol OD, ation ry of ion s viso ified ir in OBO ike C o ea peciaes Suhum (n = 30) Dormaa (n = 30) Assin 6.7 13.3 6.7 nitation 96.6 100.0 83.3 icals used 76.7 76.7 46.7 f harvest 96.6 96.6 76.7 d storage 90.0 93.3 70.0 aking 93.3 90.0 76.7 tation 83.3 90.0 76.7 rying 60.0 66.7 90.0 ata 2010–2011. farmers satisfied with selected components of COCOBOD’s price policy. component Suhum (n = 30) Dormaa (n = 30) Assi ount) 60.0 83.3 13.3 e increase 90.0 90.0 76.7 nouncement 13.3 30.0 16.7 70.0 70.0 60.0 ata 2010–2011. % of the farmers thought that the cocoa variety deter- nal quality of cocoa beans. Table 4 also shows that most nowledged the importance of farm maintenance (95%), harvesting (89%), good pod storage (83%), fermenta- and drying (71%) in enhancing the quality of the cocoa produce. They explained, however, that carrying out all ry farm practices requires extra costs in terms of time It is not always possible for farmers to meet these costs y do not have an adequate incentive to do so. formation problem n reason why the incentive is inadequate appears to be estic cocoa market suffers from a number of informa- etries. Farmers tend to have more knowledge than the OCOBOD about the production and post-harvest prac- ply and hence about some aspects of the quality of their s prior to sale. At the same time, farmers lack informa- other important aspects of bean quality and required hat is generated and shared at levels far above them in ain. metries persist partly because the LBCs and COCOBOD effective mechanisms for monitoring the production rmers. The high cost of monitoring each farmer’s activ- xplain the absence of such procedures. The only time effect. for far by farm Farme by shir 3.3.3. Fro ers it w betwee COCOB have n ity con COCOB inform Minist extens ized ad the un out the COC tions l It is to and esctices are monitored is during the government’s mass cocoa farms, where the supervisors of spraying gangs farmers have carried out their farm maintenance activ- s weed removal and pruning the trees, before the cocoa rayed against insects (capsids) and fungal disease (black the 1980s, when cocoa farmers were better organized, organizations had rules for monitoring members’ prac- ollapse of the farmer organizations throughout the ing communities arose from political interference, poor n, and growing mistrust among farmers that increased onged periods of political turbulence [35]. Though there lla cocoa farmer organization, called the Ghana Cocoa nut Farmers Association, the association is active only level. explanation for the observed information asymmetry hana’s cocoa industry the cocoa beans are not graded chase by the LBCs. It is only after the purchased cocoa ulked, sorted, and evacuated from the farms to district the strict quality control procedures of COCOBOD take that under remains hig Cocoa Offic sion agents these distri Most far BOD’s price of cocoa, t farmers the even thoug price annou farmers aga markets. W perceive th the quality 3.3.4. An in Notwith ment, farm makes them (n = 30) W. Akropong (n = 30) Total (n = 120) 0.0 6.7 100.0 95.0 70.0 67.5 86.7 89.2 80.0 83.3 100.0 90.0 86.7 84.2 70.0 71.7 (n = 30) W. Akropong (n = 30) Total (n = 120) 70.0 61.6 90.0 86.7 10.0 17.5 80.0 62.5 nwhile, because of high competition among the LBCs ’ beans, LBCs buy all the cocoa beans offered to them and recondition them later through drying and sorting. s have a high incentive to reduce their production costs some of the recommended practices. wledge problem e analyses of the focus group discussions with farm- bserved that, apart from the information asymmetries rmers and the LBCs, farmers find the linkage between olicies and cocoa bean quality hard to read. Farmers ceived training and feedback on many aspects of qual- standards. The merger of the Cocoa Service Division of which formerly was in charge of the dissemination of , with the Agricultural Extension Department of the Food and Agriculture (that became known as the unified ystem) marked the beginning of the collapse of special- ry services for cocoa farmers. Extension officers under extension system are not sufficiently resourced to carry formation task with respect to this crop [36]. D has recently partnered with private sector organiza- adbury Plc. to provide extension services to farmers. rly to assess the impact on information asymmetries lly on farmers’ knowledge. However, there is a concern this arrangement, the farmer: extension worker ratio h. For example, at the time of this study, none of the es in the districts studied had more than seven exten- . This is woefully inadequate, considering that each of cts had no less than 90 cocoa growing villages. mers interviewed were generally satisfied with COCO- policy (Table 5). Apart from the amount paid per bag he expectation of annual price increments gives the assurance that their demands are being recognized h they were mostly not comfortable with the timing of ncements. Under their price policy, COCOBOD insures inst world price volatility by selling cocoa in forward indfalls are paid to farmers as bonuses. Farmers do not e current bonuses paid to them as in any way related to of their beans. come problem standing the seemingly favourable price policy environ- ers are faced with an income problem that sometimes reluctant to invest into quality-related activities. It 12 W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 is instructive to illustrate farmers’ income problems by way of a numerical example. In line with the data from our survey and other national est hectare and hectare [14 we found, a of US$ 3.57 and recipro labour. In t stood at US ing for crop only for the Suppose three types tenant–farm enue: 250 × The acco costs - depe Owner–f 32 days), w 557.50 − 32 Tenant–f Abunu is a farmer and and landow hiring of la his output 3.57 − 1/2 × Caretake tract where even larger the output i 3.57 − 2/3 × Howeve but econom Implicit cos earned by w employmen of farmers’ 40 × US$ 3.5 land, so he h 278.57 if he 557.50 = US In sum, i to hire them earns an ec if he leas case of th (443.26–14 profit of US makes an e The exa position an ments wou Although th the exampl non-price in are unable would know This con to be makin other uses financial se risks associ alternative ers’ financia Two other key issues can be mentioned that affect the finan- cial position of farmers. First, during the interviews and focus discu f pr nnou decis ses a ittee me tobe meti us ye ncem at ti e for lway ondl s’ fin ng sc going thou duc e pre varie ble t ly be ra re ave t rme cussi stitu blem rmat m is mon ho ocola 39] p cultu reme nism o en ly im en th rela ed a cipie nd t ses. B OD, umb d by OBO y for als h OD ss-sp pray with icide m be hemiimates, suppose an annual output of 250 kg beans per an annual labour requirement of 80 working days per ,33,34]. Also suppose that, again in agreement with what bout 40% of the labour requirement is hired at the cost per day, 10% comes from non-paid sources like family cal labour; and the remaining 50% is the farmer’s own he 2010–2011 season the price per kg of cocoa beans $ 2.23. Since the government provides chemical spray- protection throughout the cocoa belt, the farmer pays cost of hired labour. a farmer crops one hectare of cocoa. There are of farmers: those who own the land (owner–farmer), er, and caretakers. Each of them earns the same rev- US$ 2.23 = US$ 557.50. unting profits - revenue minus explicit out-of-pocket nd on the type of farmer. armer. Since 40% of his labour requirement is hired (i.e., e find for his accounting profit (AP) AP (owner) = US$ × US$ 3.57 = US$ 443.26. armer. He is always tied to an Abunu land use contract. land use system in which a piece of land is given to a crops are shared on a fifty–fifty basis between farmer ner. The farmer’s explicit costs include payment for bour and, under the Abunu land use contract, half of as rent. Therefore, AP (tenant) = US$ 557.50 − 32 × US$ US$ 557.50 = US$ 164.51. r. He is always contracted under the Abusa land use con- he earns one third of the output. His explicit costs are , since under the Abusa land use contract two-thirds of s paid as rent, so AP (caretaker) = US$ 557.50 − 32 × US$ US$ 557.50 = US$ 71.59. r, the relevant income yardstick is not accounting profit ic profit, which also takes account of implicit costs. ts are captured by the wage income a farmer could have orking on someone else’s farm or in another form of t. Since 50% of the farm labour requirement consists own labour, each type of farmer has an implicit cost of 7 = US$ 142.80. Moreover, an owner could also lease his as an additional implicit cost of 1/2 × US$ 557.50 = US$ rented his land under the Abunu system, and 2/3 × US$ $ 371.67 under the Abusa system. f we assume that farmers always have the opportunity selves out or to lease their land, then an owner–farmer onomic profit of US$ 21.71 (443.26–142.80–278.75) es his land under the Abunu system or, in the e Abusa system, an economic loss of US$ 71.21 2.80–371.67). A tenant–farmer earns an economic $ 21.71 (164.51–142.80); whereas a caretaker always conomic loss of US$ 71.21 (71.59–142.80). mple illustrates the fragile basis of farmers’ economic d helps explain why making quality-related invest- ld require stronger incentives to be provided to them. e assumptions in the example may be oversimplified, e illustrates the fact that without adequate price and centives most farmers, that is, tenants and caretakers, to carry out the recommended practices, even if they what these are. straint arises from the financial losses farmers appear g because of the relatively low revenue compared with of their labour and the high cost of renting land. Poor rvices for cocoa farmers, because of the perceived high ated with lending to farmers, and because of lack of sources of income during off-season periods affect farm- l position, contribute to the income problem. group ment o were a keting purcha Comm the com and Oc and so previo annou firm th incom have a Sec farmer weighi at the that al the Pro bag, th which been a probab the ext they h from fa 4. Dis 4.1. In Pro to info proble tions a those w the ch et al. [ of agri measu mecha order t ticular betwe The describ the re BOD a proces COCOB ing a n respon COC countr chemic COCOB out ma mass-s season to pest proble right cssions it emerged that the timing of the announce- oducer prices varied from season to season and often nced too late to inform farmers’ production and mar- ions. The general schedule of events is as follows. Cocoa re halted in June each year. A Producer Price Review (PPRC) announces the new prices that will be paid at ncement of cocoa bean purchases between September r each year. Farmers begin harvesting by the end of July mes have to sell their cocoa beans at the prices of the ar between July and October because of delays in the ent of new prices. This seems on the face of it to con- ming is an issue that may lead to a considerable loss of farmers, particularly since up to this year the new prices s been higher than in the previous years. y, rent-seeking activities by competing LBCs also affect ancial position. An example is their adjustment of the ales in order to obtain more cocoa beans from farmers price. Some farmers and LBCs interviewed mentioned gh there is the so-called ‘official Accra weight’, which is er Price Review Committee’s (PPRC’s) unit of 64 kg per vailing weight is the ‘village weight’ used by the LBCs, s between 65 kg and 70 kg per bag. Farmers have not o negotiate their way out of this unfortunate position, cause they are weakly organized. The LBCs argue that venue accrued from the adjusted scales covers the risk o bear when they purchase cocoa beans of low quality rs. on tions and actor interactions s related to commodity quality are often attributed ion asymmetry [37–39]. The information asymmetry easier to understand when one analyses the interac- g the three main categories of actors in the sector i.e., ensure the movement of cocoa beans from the farm to te manufacturer–farmers, LBCs and COCOBOD. Hueth osit that when incentive problems affect the quality ral commodities, then input control, field visits, quality nt, and general price increases are the best institutional s for co-ordinating the interaction between actors in sure quality [40,41]. Two interactions seem to be par- portant here: between COCOBOD and farmers, and e LBCs and farmers. tion between COCOBOD and farmers can be best s paternalistic. The farmers consider themselves to be nts of policies, technologies and inputs from COCO- o have minimal participation in the decision-making ecause farmers do not supply cocoa beans directly to the latter relies solely on reciprocity, that is, by design- er of beneficial policies in the hope that farmers will supplying quality cocoa beans. D regulates all the chemical inputs imported into the use on cocoa farms even though sometimes unapproved ave found their way to the market for farmers’ use. Also, uses part of the export revenues from cocoa to carry raying of all cocoa farms at least twice per season. The ing campaign has been expanded since the 2008–2009 the introduction of foliar liquid fertilizers in addition s and fungicides. This helps to reduce the information cause the majority of cocoa farms are treated with the cals at least once a year. However, it does not effectively W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 13 tackle the quality aspects of cocoa pests and diseases because it is calendar-based and not need-based. It was observed during this study that t as fraudule application ules that do interference Price po COCOBOD because coc PPRC, which recommend finds it dif though farm resentative adequately Over the ye sector refor level equiva margin. Pre age of the the export (and the Ca farmers fro 2008–2009 cient motiv quality. Under th cocoa secto of testing a process wh plying low- sell high-qu ity already buyers pur two quality cocoa on t seem to ha long as the ity beans. O to different beans they not give mu icy. This is b COCOBOD. There ar incentives include an i a 45% fertili 2600 childr ible house m These po asymmetrie able to acce or they are crop. The ho for sharecro ers’ children accessing th dren. For ex village scho too poor to ships. Informa between fa since farme reforms of the 1990s, only the Produce Buying Company, a sub- sidiary of COCOBOD, purchased cocoa beans. While this monopoly me n losel lity. ean n of m illag This turn tly L s stic e far seeki ting tack ctio stitut exec tion iring omin lity b that s to the COB ason reco es. mers cted orce titut m an e re S-SIS bean udy coun rks o aris selec rem of t hen r wo coa p ease . Exp nt m olicy inst shou oup. evid ment for fa h), c r enh he b e ins antit e loche mass-spraying campaign also faces such challenges nt diversion of the approved chemicals, the inefficient techniques used by spraying gangs, application sched- not follow COCOBOD recommendations, and political . licy is used to co-ordinate the interactions between and farmers. It is effective for the sector as a whole oa supply responds positively to prices. However, the is responsible for setting cocoa producer prices (their ation is then forwarded to the Cabinet for approval), ficult to make the voices of the farmers heard even ers are represented on the PPRC [42]. The farmer rep- s interviewed indicated that they are sometimes not briefed about the methods used to determine prices. ars, two modes have been adopted. Prior to the cocoa ms in the 1990s, the policy was to set the price at a lent to the estimated cost of production plus a profit sently, price-setting is based on a calculated percent- free on board (f.o.b) price that Ghana receives from of cocoa beans. Even though the PPRC recommended binet accepted) an increase in the price paid to cocoa m 23.3% of the f.o.b price in 1983–1984 to up to 73% in , it is not known which of the two modes provides suffi- ation to farmers to want to further enhance cocoa bean e current price policy, the information problem in the r seems difficult to solve. In principle, a combination nd price differentiation could initiate a self-selection ere some farmers would be discouraged from sup- quality beans, and others encouraged to produce and ality beans. From the perspective of COCOBOD, qual- is factored into the pricing formula since its licensed chase only premium quality cocoa. Furthermore, the grades sold from Ghana are marketable as premium he international market. Hence, COCOBOD does not ve any incentive to introduce differentiated prices as y can be sure of obtaining a sufficient volume of qual- n the other hand, the LBCs may have an incentive iate prices in order to increase the volume of quality buy from farmers, yet the cocoa marketing rules do ch room for the LBCs to implement such a price pol- ecause LBCs do not receive a differentiated price from e other policies introduced by COCOBOD that provide to cocoa farmers to improve their production. These nput credit programme (the so-called hi-tech scheme), zer subsidy, annual scholarship grants awarded to about en of cocoa farmers and the staff of COCOBOD, and flex- ortgage schemes. licies, however, do not help to reduce the information s in the sector. Also, the majority of farmers are not ss these benefits because they either are smallholders sharecroppers who do not own the cocoa farms they use mortgage scheme, for instance, still is too expensive ppers and, apart from the fact that only 40% of farm- benefit from the scholarships, some of the criteria for e cocoa scholarships do not favour smallholders’ chil- ample, the examination results of children attending ols i.e., those attended by farmers’ children, are often meet the pass mark required to access cocoa scholar- tion asymmetry is more persistent in the interaction rmers and LBCs than between farmers and COCOBOD rs supply their beans straight to LBCs. Prior to the had so were c for qua cocoa b ductio of the v ment. that in Presen farmer over th profit- marke dently taken a 4.2. In The interac of acqu shortc of qua shown farmer reduce and CO real re ing the practic Far undete to enf for ins proble nerativ the Co cocoa This st into ac netwo gestion of self- price p impact tor and Anothe the co to incr quality differe price p The paper tion Gr require experi prices researc gies fo from t provid and qu into thegative socio-economic effects, farmers’ cocoa beans y inspected at the village-level and a premium was paid This partly explains the high quality status Ghanaian s enjoyed in the years preceding the reforms. The intro- ore LBCs after the reforms brought with it a weakening e-level mechanism for bean quality checks and enforce- has provided room for rent-seeking activities by LBCs undermine the grounds for paying quality premiums. BCs have little or no mechanisms for ensuring that k to recommended practices. The fundamental concern mer–LBC interaction is that LBCs are becoming merely ng agents of COCOBOD. Since the rules regarding the of cocoa beans give the LBCs little room to indepen- le the information problem, they themselves have not n to remedy the situation. ional gaps and opportunities for experimentation ution of a diagnostic study on the institutions governing s in a public-interest sector like cocoa faces the problem adequate data. Moreover, it is difficult to point out any gs while Ghana enjoys a high reputation for the export eans. Notwithstanding these difficulties, the study has COCOBOD’s policies have provided some incentives to enhance quality but also that these policies only partly information asymmetries among cocoa farmers, LBCs, OD. It has been demonstrated that this may be the why some farmers shirk the responsibility of adopt- mmended production and post-harvest management who do not stick to the recommended practices remain and unsanctioned: no mechanism nor procedure exists their adoption. These gaps represent opportunities ional change that could help reduce the information d at the same time provide cocoa farmers a more remu- ward for their activities. The original entry point of cocoa domain research was to sustain the quality of by means of changes in the structure of incentives. suggests that further research is warranted that takes t the potential influence of farmer organizations and n member farmers’ production practices. The main sug- ing from this study is that the appropriate development tion mechanisms, such as quality testing coupled with iums paid at farm gate, could overcome the negative he existing information asymmetries in the cocoa sec- ce is an option worth exploring as a policy experiment. rthwhile institutional experiment would be to make rice policy formulation process more transparent and the transparency of the relation between pricing and loration of the economic, social and quality impacts of odes of pricing would also provide useful insight into options for enhancing cocoa bean quality. itutional experiments suggested by the analysis in this ld be of interest to the cocoa Concertation and Innova- However, the higher level institutional changes would ence and feedback from the farmer- and village-level s. Experiments that provide differentiated incentive rmers (which might become part of our on-going (PhD) ould provide information about cost-effective strate- ancing cocoa bean quality by raising standards right eginning of the value chain. Such experiments would ight into how farmers might respond, in terms of quality y, to the introduction of alternative pricing mechanisms al cocoa market. 14 W. Quarmine et al. / NJAS - Wageningen Journal of Life Sciences 60– 63 (2012) 7– 14 5. Conclusions This pap in the coco that act as a beans. The ing socio-ec the process tive ease of internation The ana only premiu could be inc ity of their are unable o farm practi The key the interact ized by pro monitoring organizatio more, our off-season t and have li tices or COC The curr the incentiv experiment tutional im enhance an that farmer Acknowled This stu Strengthen We are grat culture at S assistance o for helping References [1] P. Jano, D constrain sented at [2] T. Wahyu Presented (IPiCEX), [3] T. William tute of De [4] D.A. Kote vegetable [5] K. Kpodo Newswir [6] A. Laven, London, 2 [7] I. Osei, Su spective, Indonesia [8] E.N.A. Do profitabil A. Youde Smallhold [9] F.W. Gee cesses: a 1257–127 [10] F.W. Gee Insights a Research [11] D. Hounk approach Enhancing Innovation Systems Performance in Smallholder African Agricul- ture, Elmina, Ghana, 2009, pp. 11–14. [12] R. Kemp, J. Schot, R. Hoogma, Regime shifts to sustainability through processes iche f lysis ip, R. ice an beny cial fo 3. Asens The L shing aah, H ocoa . Bate ana’s C Worl akane anaian oodh ernati dquis C. Edq Hall, L tic as er 20 . Oluy logica –614 eeuw otiati sented Devel sumin g sust pp. . Pouls quali and, T 9. . Sukh Revita Optio Toba Gilmo h Inte onte, P ue cha reisin elopm icy Re Anim the q ociati . Dong ers p p Scie –302 eal, A. ra, 20 poku cquay ductio Ton, G econ oire, a & Pol .A. Do ment genin Khera tural p 1. . Aker sm, Qu ueth, tracts tural E Laffon h ed., olmst iew 8 Vigne ieve: ernatier reports the findings of a diagnostic study on quality a sector in Ghana, focusing on the institutional factors disincentive for farmers to enhance the quality of cocoa concept of cocoa bean quality is complex, encompass- onomic, physical, and bio-chemical aspects as well as of cocoa production. Currently, because of the rela- observing physical characteristics, both domestic and al markets emphasize physical quality. lysis highlights the fact that although Ghana exports m quality cocoa beans, the volume of high quality beans reased if farmers were motivated to enhance the qual- harvested cocoa beans. At present, farmers sometimes r unwilling to invest resources into the recommended ces because there are little or no incentives to do so. explanation identified for the lack of incentives is that ion among farmers, LBCs and COCOBOD is character- blems of information asymmetry. The lack of farm and grading services, and the lack of strong farmer ns, allow the information problem to persist. Further- analysis shows that farmers face an income problem hat restricts the potential for investment of any kind, ttle knowledge of either the recommended farm prac- OBOD policies and standards with respect to quality. ent policies of COCOBOD do not address adequately e problem. The policy gap presents an opportunity for ing at farmer- and higher levels that might lead to insti- provements through the value chain that would help d sustain cocoa bean quality and increase the reward s receive. gements dy was funded by the Convergence of Sciences – ing Innovations Systems (CoS-SIS) research programme. eful to Mr. Felix Awu, MIS officer at the Ministry of Agri- uhum Craboar-Coaltar District. We also appreciate the f the staff of COCOBOD, in all the districts we visited, us to complete this study. . Mainville, The cocoa marketing chain in ecuador: analysis of chain ts to the development of markets for high-quality cacao, Paper Pre- IAMA Conference, Parma, Italy, 2007. di, The world scenario of cocoa production and consumption, Paper at 2nd International Plantation Industry Conference and Exhibition UiTM Shah Alam, Selangor, Malaysia, 2008. s, An African success story: Ghana’s cocoa marketing system, Insti- velopment Studies Working Paper 318, 2009. y, W.S.K. Gbewonyo, K. Afreh-Nuamah, High chlorpyrifos levels on s in Ghana, Pesticides News 80 (2008) 4–6. , Ghana Light Crop Cocoa Purchases Pull Ahead of 06/7, Dow Jones es, New York, 2006. Marketing Reforms in Ghana’s Cocoa Sector, ODI Background Note, 007. stainable practices in the global cocoa economy—a producer’s per- Paper Presented at 4th Indonesia International Cocoa Conference, , 2007. rmon, O. Sakyi-Dawson, Exploring opportunities for enhancing the ity and sustainability of cocoa production in Ghana, in: A. van Huis, owei (Eds.), Towards Enhancing Innovation Systems Performance in er African Agriculture, Elmina, Ghana, 2009, pp. 68–69. ls, Technological transitions as evolutionary reconfiguration pro- multi-level perspective and a case-study, Research Policy 21 (2002) 4. ls, From sectoral systems of innovation to socio-technical systems bout dynamics and change from sociology and institutional theory, Policy 33 (2004) 897–920. onnou, A. van Huis, A. Röling, B. Sterk, The innovation systems proposed for CoS-SIS, in: A. van Huis, A. Youdeowei (Eds.), Towards of n Ana [13] A. R Cho [14] O. A spe 200 [15] W. in: Wa [16] F. B of c [17] M.J Gh The [18] T. T Gh [19] J. W Int [20] C. E in: [21] A.J. nos Pap [22] K.A eco 609 [23] C. L neg Pre on [24] S. A pin 96 [25] C.S and [26] R. D 199 [27] D.A in: and and [28] M. 16t [29] S. P val [30] C. B dev Pol [31] M. ing Ass [32] L.N farm Cro 299 [33] F. T Acc [34] K. O S. A Pro [35] G. and d’Iv egy [36] E.N age Wa [37] M. cul 200 [38] G.A ani [39] B. H con cul [40] J.J. 12t [41] B. H Rev [42] M. Ach Intormation: the approach of strategic niche management, Technology and Strategic Management 10 (1998) 175–196. Kemp, Technological change, in: S. Rayner, E.L. Malone (Eds.), Human d Climate Change, Battelle Press, Columbus, OH, 1998, pp. 327–399. ega, J. Gockowski, Labour practices in the cocoa sector of Ghana: with a cus on the role of children, STCP/IITA Monograph, IITA, Ibadan, Nigeria, o-Okyere, The response of farmers to Ghana’s adjustment policies, ong-Term Perspective Study of Sub-Saharan Africa, The World Bank, ton, DC, 1990, pp. 74–83. arnessing farmer associations as channels for enhanced management holdings in Ghana, Scientific Research and Essay 3 (2008) 395–400. man, A. Meeraus, D.M. Newbery, W. Asenso-Okyere, G.T. O’Mara, ocoa Pricing Policy, Agricultural and Rural Development Department, d Bank, Washington, DC, 1990. , Incentives embedded in institutions: the case of share contracts in cocoa production, The Developing Economies 38 (2000) 374–397. ill, Institutional change—a framework for analysis, in: Wageningen onal Occasional Paper, Wageningen International, Wageningen, 2008. t, B. Johnson, Institutions and organizations in systems of innovation, uist (Ed.), Systems of Innovation, Pinter, London, 1997, pp. 41–63. .K. Mytelka, B. Oyelaran-Oyeyinka, Concepts and guidelines for diag- sessments of agricultural innovation capacity, UNU-MERIT Working 06-017, 2006. ole, The effect of weather on cocoa production in different agro- l zones in Nigeria, World Journal of Agricultural Sciences 6 (2010) . is, Innovation as a process of network building, social learning and on—implications for transdisciplinary research collaboration, Paper at Trans-KARST 2004—International Transdisciplinary Conference opment and Conservation of Karst Regions, Hanoi, 2004. g-Brempong, D.B. Sarpong, N.A. Anyidoho, R. Kaplinsky, J. Leavy, Map- ainable production in Ghanaian cocoa. Report to Cadbury, Accra, 2008, en, H.J. Juhl, K. Kristensen, A.C. Bech, E. Engelund, Quality guidance ty formation, Food Quality and Preferences 7 (1996) 127–135. he International Cocoa Trade, Woodhead Publishing Ltd., Cambridge, a, Potential value added products from Trinidad and Tobago cocoa, lisation of the Trinidad and Tobago cocoa industry—Targets, Problems ns, The Association of Professional Agricultural Scientists of Trinidad go, Trinidad and Tobago, 2003, pp. 69–73. ur, Quality and food safety in a sustainable cocoa supply chain, in: rnational Cocoa Research Conference, Bali, 2009. . Gibbon, Quality standards, conventions and the governance of global ins, Economy and Society 34 (2005) 1–31. ger, X. Diao, S. Kolavalli, J. Thurlow, The role of cocoa in Ghana’s future ent, in: Ghana Strategy Support Program (GSSP), International Food search Institute, Accra, 2008, 24 pp. -Kwapong, R. Appiah, K. Osei-Bonsu, A review of the factors affect- uality of processed cocoa beans in Ghana, Journal of Ghana Science on 9 (2007) 103–119. o, E.O. Aigbekaen, C.O. Jayeola, L.A. Emaku, S.B. Orisajo, Influence of ractices on cocoa bean quality: Nigeria field experience, in: African nces Conference, African Crop Sciences Society, Uganda, 2009, pp. . Zeitlin, H. Hammah, Ghana Cocoa Farmers Survey, Ghana Cocoa Board, 06. -Ameyaw, F. Baah, V. Anchirinah, H.K. Dzahini-Obiatey, R.A. Cudjoe, e, Y.S. Opoku, Cocoa Manual—A Source Book for Sustainable Cocoa n, Cocoa Research Institute of Ghana, Tafo, 2010. . Hagelaar, A. Laven, S. Vellema, Chain governance, sector policies omic sustainability in cocoa: a comparative analysis of Ghana, Côte nd Ecuador, in: Markets, Chains and Sustainable Development Strat- icy Papers, Stichting DLO, 2008. rmon, From a technology focus to innovation development: the man- of cocoa pests and diseases in Ghana, CERES, Wageningen University, gen, 2006. llah, J. Kirsten, The new institutional economics: applications for agri- olicy research in developing countries, MSSD Discussion Papers 41, lof, The market for ‘lemons’: quality uncertainty and the market mech- arterly Journal of Economics 84 (1970) 488–500. E. Ligon, S. Wolf, S. Wu, Incentive instruments in fruit and vegetable : input control, monitoring, measuring, and price risk, Review of Agri- conomics 21 (1999) 374–389. t, D. Martimort, Theory of Incentives – The Principal – Agent Model, Princeton University Press, Princeton, NJ, 2002. röm, P. Milgrom, The firm as an incentive system, American Economic 4 (1994) 972–991. ri, P. Santos, What Does Liberalization Without Price Competition The Case of Cocoa in Ghana, Ghana Strategy Support Program (GSSP), onal Food Policy Research Institute (IFPRI), 2008.